Method for detecting defect of images and electronic device

An electronic device comprising:

• a processor;
• and
• a non-transitory storage medium coupled to the processor and configured to store a plurality of instructions, which cause the processor to:
• input flawless sample training images into an autoencoder of a defect detection model, and calculate first latent feature of the flawless sample training images by a coding layer of the autoencoder;
• input the first latent feature into a decoding layer of the autoencoder, and calculate first reconstructed images of the flawless sample training images, and calculate a first reconstruction error between the flawless sample training images and the first reconstructed images by a first preset error function;
• input the flawless sample training images and the first reconstructed images into a discriminator of the defect detection model, train the discriminator, and calculate an adversarial learning error between the discriminator and the autoencoder by a second preset error function;
• calculate a sample error according to the first reconstruction error and the adversarial learning error;
• determine an error threshold based on the sample error in response that the sample error satisfies a first preset condition;
• obtain testing sample images, input the testing sample images into the coding layer of the autoencoder after training, and calculate second latent feature of the testing sample images, and input the second latent feature into the decoding layer of the autoencoder after training, and calculate the second reconstructed images of the testing sample images, and calculate a difference between the testing sample images and the second reconstructed images by the first preset error function, and take the difference as a second reconstruction error;
• determine a detection result of the testing sample images according to the second reconstruction error and the error threshold.

A non-transitory storage medium having stored thereon instructions that, in response that executed by at least one processor of an electronic device, causes the least one processor to execute instructions of a method for detecting defect of images, the method comprising:

• inputting flawless sample training images into an autoencoder of a defect detection model, and calculating first latent feature of the flawless sample training images by a coding layer of the autoencoder;
• inputting the first latent feature into a decoding layer of the autoencoder, and calculating first reconstructed images of the flawless sample training images, and calculating a first reconstruction error between the flawless sample training images and the first reconstructed images by a first preset error function;
• inputting the flawless sample training images and the first reconstructed images into a discriminator of the defect detection model, training the discriminator, and calculating an adversarial learning error between the discriminator and the autoencoder by a second preset error function;
• calculating a sample error according to the first reconstruction error and the adversarial learning error;
• determining an error threshold based on the sample error in response that the sample error satisfies a first preset condition;
• obtaining testing sample images, inputting the testing sample images into the coding layer of the autoencoder after training, and calculating second latent feature of the testing sample images, and inputting the second latent feature into the decoding layer of the autoencoder after training, and calculating the second reconstructed images of the testing sample images, and calculating a difference between the testing sample images and the second reconstructed images by the first preset error function, and taking the difference as a second reconstruction error;
• determining a detection result of the testing sample images according to the second reconstruction error and the error threshold.